5

Assuming we're dealing with an artificial neural network (e.g. a ConvNet), which was trained by a large dataset of human faces.

Are there any known issues or challenges where facial recognition would fail? I'm not talking about covering half of the face, but some simple common things, such as wearing glasses, a hat, jewelry, having face painting, or tattoos. Can this successfully prevent AI from recognizing the face? If so, what are the current methods for dealing with such challenges?

nbro
  • 39,006
  • 12
  • 98
  • 176
kenorb
  • 10,423
  • 3
  • 43
  • 91
  • Wearing a mask which consists of someone else's face will fool the algorithm unless it's trained for such a thing. – Pimgd Aug 09 '16 at 14:32

1 Answers1

4

Facial recognition works by essentially turning your face into a point cloud, recognizing eyes, cheeks, nose, mouth, etc.

Unfortunately it doesn't look at the top of your head (hair is very hard to differentiate from other hair and doesn't have many features). Face paintings would be your best bet since they can be easily changed, tattoos not so much. Once somebody has a photo of your face with your tattoo on it, you're busted. Glasses will work if they're opaque and hide your eyes (sunglasses). The facial recognition software does not recognize jewelry, as it's tiny, and very easy to remove and put on. Ideally you want to have your face professionally made up with makeup and fake skin (basically a fake face).

Some challenges can be anything like an unclear picture, a picture from the wrong angle (which can probably be mathematically calculated and restructured), or like you said, face paintings to hide facial features.

Taylor Swift
  • 56
  • 1